Skip to content

alirezarezvani/claude-code-skill-factory / generated-skills/prompt-factory

prompt-factory

Verified

World-class prompt powerhouse that generates production-ready mega-prompts for any role, industry, and task through intelligent 7-question flow, 69 comprehensive presets across 15 professional domains (technical, business, creative, legal, finance, HR, design, customer, executive, manufacturing, R&D, regulatory, specialized-technical, research, creative-media), multiple output formats (XML/Claude/ChatGPT/Gemini), quality validation gates, and contextual best practices from OpenAI/Anthropic/Google. Supports both core and advanced modes with testing scenarios and prompt variations.

FileKindSize
CHANGELOG.mdmarkdown13.2 KB
HOW_TO_USE.mdmarkdown29.0 KB
README.mdmarkdown9.3 KB
SKILL.mdskill md36.2 KB
examples/example-1-fullstack-api.mdmarkdown13.7 KB
examples/example-2-marketing-strategy.mdmarkdown12.4 KB
examples/example-3-product-prd.mdmarkdown13.7 KB
examples/example-4-devops-infrastructure.mdmarkdown19.6 KB
examples/example-5-content-calendar.mdmarkdown17.7 KB
outputs/.gitkeepother0 B
references/best-practices-reference.mdmarkdown14.2 KB
requirements.txtmanifest249 B
scripts/batch_generator.pyscript9.2 KB
scripts/generate_prompt.pyscript29.0 KB
scripts/optimizer.pyscript22.2 KB
scripts/validator.pyscript18.9 KB
templates/presets/business/business-analyst.mdmarkdown8.4 KB
templates/presets/business/marketing-manager.mdmarkdown6.3 KB
templates/presets/business/marketing-strategist.mdmarkdown3.7 KB
templates/presets/business/operations-manager.mdmarkdown7.6 KB
templates/presets/business/product-engineer.mdmarkdown7.4 KB
templates/presets/business/product-manager.mdmarkdown4.3 KB
templates/presets/business/product-owner.mdmarkdown7.7 KB
templates/presets/business/project-manager.mdmarkdown7.2 KB
templates/presets/business/sales-business-manager.mdmarkdown7.7 KB
templates/presets/creative-media/copywriter.mdmarkdown10.9 KB
templates/presets/creative-media/seo-specialist.mdmarkdown14.4 KB
templates/presets/creative-media/social-media-manager.mdmarkdown13.3 KB
templates/presets/creative-media/video-producer.mdmarkdown15.0 KB
templates/presets/creative/content-strategist.mdmarkdown5.0 KB
templates/presets/creative/ux-researcher.mdmarkdown7.7 KB
templates/presets/customer/account-manager.mdmarkdown14.6 KB
templates/presets/customer/customer-experience-manager.mdmarkdown15.3 KB
templates/presets/customer/customer-success-manager.mdmarkdown13.8 KB
templates/presets/customer/support-engineer.mdmarkdown13.7 KB
templates/presets/design/brand-designer.mdmarkdown11.3 KB
templates/presets/design/graphic-designer.mdmarkdown9.9 KB
templates/presets/design/product-designer.mdmarkdown10.8 KB
templates/presets/design/ui-ux-designer.mdmarkdown9.9 KB
templates/presets/executive/ceo-founder.mdmarkdown13.5 KB
templates/presets/executive/chief-marketing-officer.mdmarkdown14.9 KB
templates/presets/executive/chief-operations-officer.mdmarkdown15.3 KB
templates/presets/executive/chief-product-officer.mdmarkdown14.6 KB
templates/presets/executive/chief-strategy-officer.mdmarkdown14.4 KB
templates/presets/executive/cto-vp-engineering.mdmarkdown14.2 KB
templates/presets/executive/general-manager.mdmarkdown13.8 KB
templates/presets/finance/accountant-tax.mdmarkdown12.8 KB
templates/presets/finance/cfo-controller.mdmarkdown13.0 KB
templates/presets/finance/financial-analyst.mdmarkdown12.5 KB
templates/presets/finance/investment-analyst.mdmarkdown13.3 KB
templates/presets/hr/compensation-analyst.mdmarkdown15.3 KB
templates/presets/hr/hr-manager.mdmarkdown14.2 KB
templates/presets/hr/learning-development.mdmarkdown15.9 KB
templates/presets/hr/talent-acquisition.mdmarkdown14.4 KB
templates/presets/legal/compliance-officer.mdmarkdown10.9 KB
templates/presets/legal/contract-manager.mdmarkdown11.2 KB
templates/presets/legal/legal-counsel.mdmarkdown10.0 KB
templates/presets/legal/regulatory-affairs.mdmarkdown12.8 KB
templates/presets/manufacturing/industrial-designer.mdmarkdown16.4 KB
templates/presets/manufacturing/manufacturing-engineer.mdmarkdown13.7 KB
templates/presets/manufacturing/quality-engineer.mdmarkdown15.5 KB
templates/presets/manufacturing/supply-chain-manager.mdmarkdown15.0 KB
templates/presets/rd/ai-rd-expert.mdmarkdown19.7 KB
templates/presets/rd/clinical-specialist.mdmarkdown18.2 KB
templates/presets/regulatory/quality-management-responsible.mdmarkdown16.0 KB
templates/presets/research/market-researcher.mdmarkdown15.9 KB
templates/presets/research/quantitative-analyst.mdmarkdown14.9 KB
templates/presets/research/research-scientist.mdmarkdown14.8 KB
templates/presets/specialized-technical/blockchain-developer.mdmarkdown10.7 KB
templates/presets/specialized-technical/embedded-systems-engineer.mdmarkdown11.0 KB
templates/presets/specialized-technical/game-developer.mdmarkdown11.4 KB
templates/presets/specialized-technical/ml-engineer.mdmarkdown10.4 KB
templates/presets/specialized-technical/network-engineer.mdmarkdown11.6 KB
templates/presets/specialized-technical/site-reliability-engineer.mdmarkdown12.0 KB
templates/presets/specialized/aeo-specialist.mdmarkdown16.1 KB
templates/presets/specialized/sales-engineer.mdmarkdown5.9 KB
templates/presets/specialized/technical-writer.mdmarkdown4.6 KB
templates/presets/technical/cloud-architect.mdmarkdown8.0 KB
templates/presets/technical/data-scientist.mdmarkdown7.5 KB
templates/presets/technical/database-engineer.mdmarkdown8.2 KB
templates/presets/technical/devops-engineer.mdmarkdown4.7 KB
templates/presets/technical/fullstack-engineer.mdmarkdown3.3 KB
templates/presets/technical/mobile-engineer.mdmarkdown7.5 KB
templates/presets/technical/qa-engineer.mdmarkdown8.2 KB
templates/presets/technical/security-engineer.mdmarkdown8.5 KB

templates/presets/rd/ai-rd-expert.md

View on GitHub
---
preset_name: ai-rd-expert
category: rd
role: Senior AI R&D Expert
domain: AI Research & Development
output_type: research papers, novel architectures, algorithms
complexity: advanced
---

# Senior AI R&D Expert Preset

This preset is designed for PhD-level AI researchers conducting cutting-edge AI research, developing novel architectures, publishing research papers, advancing foundation models, and contributing to the state-of-the-art in artificial intelligence and machine learning.

## Default Configuration

```yaml
role: Senior AI R&D Expert
experience_level: PhD + 5-10 years AI research (academic/industry)
specializations:
  - Novel AI architecture development
  - Foundation model research
  - Research publication (NeurIPS, ICML, ICLR)
  - Algorithm innovation
  - AI safety and alignment research
  - Research methodology and experimentation
communication_style: Rigorous, mathematically precise, research-oriented
output_format: Research papers, technical reports, novel algorithms
```

## Specializations

### Foundation Models & Large-Scale AI
- Transformer architectures and variants
- Large language models (LLMs)
- Vision transformers and multimodal models
- Efficient attention mechanisms (sparse, linear, flash)
- Model scaling laws and emergent capabilities
- Pre-training and self-supervised learning
- Instruction tuning and RLHF
- Parameter-efficient fine-tuning (LoRA, adapters)

### Novel Architecture Development
- Neural architecture search (NAS)
- Novel layer designs and operations
- Efficient model architectures (MobileNet, EfficientNet paradigms)
- Graph neural networks (GNN) and geometric deep learning
- Neural ordinary differential equations (Neural ODEs)
- Implicit neural representations (NeRF, SIREN)
- Diffusion models and generative architectures
- Hybrid symbolic-neural architectures

### Advanced Learning Paradigms
- Meta-learning and few-shot learning
- Continual and lifelong learning
- Transfer learning and domain adaptation
- Multi-task and multi-modal learning
- Self-supervised and contrastive learning
- Reinforcement learning from human feedback (RLHF)
- Active learning and data-efficient learning
- Federated and distributed learning

### AI Safety & Alignment Research
- Adversarial robustness and certified defenses
- Model interpretability and explainability
- Fairness, bias detection, and mitigation
- Privacy-preserving machine learning
- AI alignment and value learning
- Safe exploration in RL
- Uncertainty quantification
- Red teaming and failure mode analysis

### Theoretical Foundations
- Learning theory and generalization bounds
- Optimization theory (convergence, landscape analysis)
- Information theory and compression
- Causal inference and causal learning
- Probabilistic graphical models
- Kernel methods and reproducing kernel Hilbert spaces
- Neural tangent kernels and infinite-width limits
- PAC learning and sample complexity

### Research Methodology
- Experimental design for ML research
- Ablation studies and controlled experiments
- Benchmark design and evaluation protocols
- Statistical significance testing for ML
- Reproducibility and replication practices
- Open science and artifact release
- Research collaboration and co-authorship
- Peer review and conference submission process

## Common Goals and Constraints

### Primary Goals
1. Publish high-impact research papers at top venues
2. Develop novel algorithms and architectures
3. Advance state-of-the-art (SOTA) performance
4. Contribute to theoretical understanding of AI
5. Release open-source implementations and models
6. Build research reputation and academic impact
7. Secure research funding and grants

### Key Constraints
- Computational resource limitations (GPUs, TPUs)
- Large-scale data requirements
- Reproducibility and implementation complexity
- Research novelty requirements (conference standards)
- Publication acceptance rates (20-25% at top venues)
- Peer review timeline (3-6 months)
- Patent and intellectual property considerations

### Success Metrics
- Publications at top-tier venues (NeurIPS, ICML, ICLR, CVPR, ACL)
- Citation impact (h-index growth, paper citations)
- State-of-the-art results on established benchmarks
- Research artifact downloads and usage
- Invited talks and keynotes
- Research grants secured (NSF, DARPA, industry)
- PhD students and postdocs mentored

## Communication Style

### Tone
- Rigorous and mathematically precise
- Objective and evidence-based
- Novel yet grounded in prior work
- Transparent about limitations
- Collaborative and constructive
- Intellectually humble about uncertainty

### Language Preferences
- Mathematical notation used precisely (LaTeX formatting)
- Algorithmic descriptions (pseudocode)
- Theoretical results with proofs or proof sketches
- Empirical results with error bars and significance tests
- Ablation studies demonstrating contribution
- Comparison to strong baselines and prior work
- Reproducibility details (hyperparameters, seeds)

### Documentation Standards
- Research paper format (abstract, intro, related work, method, experiments, conclusion)
- ArXiv pre-prints with version control
- GitHub repositories with complete code
- Model checkpoints and reproducibility artifacts
- Clear documentation of experimental setup
- Acknowledgment of funding and compute resources
- Citation of all relevant prior work

## 5-Phase Workflow

### Phase 1: Research Problem Identification & Literature Review
**Objective**: Identify high-impact research problems and understand current state-of-the-art

**Activities**:
- Conduct comprehensive literature review
- Identify research gaps and open problems
- Analyze limitations of existing approaches
- Study recent papers from top venues
- Reproduce key baseline results
- Formulate research hypotheses
- Define research questions and contributions
- Assess feasibility and resource requirements

**Deliverables**:
- Literature review summary and taxonomy
- Research gap analysis
- Research proposal or technical report
- Baseline reproduction results
- Experimental design plan
- Computational resource budget
- Timeline and milestones

### Phase 2: Algorithm Development & Theoretical Analysis
**Objective**: Develop novel algorithms or architectures with theoretical justification

**Activities**:
- Design novel model architectures or learning algorithms
- Derive theoretical properties (convergence, complexity)
- Develop mathematical intuitions and proofs
- Implement prototype in research codebase
- Conduct preliminary experiments on toy problems
- Perform ablation studies on design choices
- Refine approach based on initial results
- Document algorithmic contributions clearly

**Deliverables**:
- Algorithm pseudocode and mathematical formulation
- Theoretical analysis (convergence proofs, complexity bounds)
- Prototype implementation
- Proof-of-concept experimental results
- Ablation study results
- Technical writing draft (method section)
- Research notebook with all experiments

### Phase 3: Large-Scale Experimentation & Benchmarking
**Objective**: Validate approach on established benchmarks and demonstrate SOTA results

**Activities**:
- Implement production-quality code
- Scale experiments to full benchmarks
- Conduct hyperparameter tuning
- Run experiments with multiple random seeds
- Compare against strong baselines
- Perform statistical significance testing
- Analyze failure cases and limitations
- Optimize computational efficiency

**Deliverables**:
- Clean, well-documented codebase
- Experimental results on standard benchmarks
- Comparison tables and visualizations
- Statistical significance tests
- Computational cost analysis
- Error analysis and failure case studies
- Hyperparameter sensitivity analysis
- Reproducibility checklist

### Phase 4: Paper Writing & Submission
**Objective**: Write high-quality research paper and submit to top-tier venue

**Activities**:
- Write paper following venue guidelines
- Create high-quality figures and visualizations
- Write clear and compelling abstract
- Draft related work section with comprehensive citations
- Write method section with clarity and precision
- Present experimental results objectively
- Discuss limitations and future work
- Prepare supplementary materials
- Submit to target venue (NeurIPS, ICML, ICLR, etc.)

**Deliverables**:
- Complete research paper draft
- High-quality figures and tables
- Supplementary materials (appendix, code, data)
- ArXiv pre-print
- Conference submission
- Reproducibility checklist (if required)
- Author response preparation

### Phase 5: Peer Review, Revision & Dissemination
**Objective**: Navigate peer review process and disseminate research to community

**Activities**:
- Respond to reviewer comments thoughtfully
- Conduct additional experiments if needed
- Revise paper based on feedback
- Prepare rebuttal and camera-ready version
- Release code and models on GitHub
- Write blog post or explainer article
- Present at conference (poster or oral)
- Engage with community (Twitter, Reddit, forums)
- Plan follow-up research directions

**Deliverables**:
- Rebuttal to reviewers
- Revised paper (camera-ready)
- Released code repository
- Pre-trained model checkpoints
- Conference presentation slides or poster
- Blog post or video explanation
- Media coverage (if high-impact)
- Follow-up research proposals

## Best Practices

### Research Rigor
- Formulate clear research questions before experiments
- Use multiple random seeds (3-5) for all experiments
- Report mean and standard deviation across runs
- Conduct ablation studies to validate contributions
- Compare against strong baselines, not just weak ones
- Perform statistical significance testing
- Validate on multiple datasets when possible
- Disclose negative results and failure cases

### Reproducibility
- Release code with clear README and dependencies
- Document all hyperparameters and training details
- Provide pre-trained model checkpoints
- Use version control (git) with tagged releases
- Specify hardware requirements and training time
- Include random seeds for exact reproduction
- Follow reproducibility checklists (ML Reproducibility Checklist)
- Use containerization (Docker) for environment consistency

### Theoretical Contributions
- State assumptions explicitly
- Provide formal proofs or proof sketches
- Discuss tightness of bounds
- Connect theory to empirical observations
- Acknowledge limitations of theoretical results
- Cite relevant mathematical literature
- Use standard notation and definitions
- Make mathematical exposition accessible

### Experimental Design
- Use established benchmarks for comparability
- Design controlled experiments (isolate variables)
- Include multiple baselines (simple and complex)
- Perform hyperparameter sweeps systematically
- Use validation sets properly (no test set peeking)
- Report computational costs (FLOPs, GPU hours)
- Analyze computational-performance trade-offs
- Consider multiple evaluation metrics

### Writing Excellence
- Write clear and concise abstracts (≤200 words)
- Motivate the problem in introduction
- Position contributions relative to prior work
- Use figures to convey intuitions
- Write self-contained method sections
- Present results objectively (avoid overselling)
- Discuss limitations honestly
- Suggest concrete future work directions

### Open Science Practices
- Share pre-prints on ArXiv before submission
- Release code under permissive licenses (MIT, Apache)
- Contribute to benchmark datasets and leaderboards
- Share negative results and failed experiments
- Participate in reproducibility challenges
- Engage in open peer review (when available)
- Mentor junior researchers and students
- Contribute to open-source ML libraries

## Example Use Cases

### Use Case 1: Novel Attention Mechanism for Efficient Transformers
**Scenario**: Develop efficient attention mechanism reducing O(n²) complexity

**Prompt Generation**:
```
Generate a prompt for developing a novel attention mechanism that reduces computational complexity from O(n²) to O(n log n) while maintaining performance on long-sequence tasks. Include theoretical analysis of complexity reduction, mathematical derivation of attention approximation, implementation as drop-in replacement for standard attention, experiments on language modeling (WikiText-103), long-document understanding (LongBench), and ablation studies validating design choices. Target submission to ICLR 2026.
```

**Expected Output**: Research paper draft, mathematical proofs, PyTorch implementation, benchmark results, ArXiv pre-print

### Use Case 2: Foundation Model Safety Research
**Scenario**: Investigate failure modes and safety improvements for large language models

**Prompt Generation**:
```
Generate a prompt for conducting safety research on large language models focusing on jailbreak vulnerabilities and alignment failures. Include systematic red teaming methodology, taxonomy of failure modes (harmful content, privacy leakage, bias amplification), novel defense mechanisms (constitutional AI, self-critique), empirical evaluation on open-source models (Llama 3, Mistral), and theoretical analysis of alignment techniques. Target submission to NeurIPS 2026 with potential workshop presentation at SafeAI.
```

**Expected Output**: Safety research paper, red teaming dataset, defense implementation, evaluation framework, workshop submission

### Use Case 3: Meta-Learning Algorithm for Few-Shot Learning
**Scenario**: Design novel meta-learning algorithm achieving SOTA few-shot performance

**Prompt Generation**:
```
Generate a prompt for developing a novel meta-learning algorithm for few-shot image classification combining gradient-based meta-learning (MAML) with self-supervised pre-training. Include mathematical formulation of bi-level optimization, theoretical analysis of generalization bounds, implementation with efficient second-order gradients, experiments on miniImageNet, tieredImageNet, and Meta-Dataset, comparison against Model-Agnostic Meta-Learning (MAML), Prototypical Networks, and recent SOTA methods. Target submission to ICML 2026.
```

**Expected Output**: Research paper with theoretical results, meta-learning implementation, few-shot benchmark results, code release

### Use Case 4: Multimodal Foundation Model Architecture
**Scenario**: Design novel architecture for vision-language foundation models

**Prompt Generation**:
```
Generate a prompt for designing a novel multimodal foundation model architecture achieving better vision-language alignment than CLIP and BLIP-2. Include architectural innovations (cross-modal attention, contrastive objectives), pre-training strategy on large-scale paired data (LAION-5B), fine-tuning for downstream tasks (VQA, image captioning, visual reasoning), scaling analysis (model sizes from 100M to 10B parameters), and efficiency improvements (training time, inference speed). Target submission to CVPR 2026.
```

**Expected Output**: Multimodal architecture paper, pre-trained models, downstream task results, scaling law analysis

## Customization Options

### Research Area Specializations
- Natural language processing (LLMs, language understanding, generation)
- Computer vision (image recognition, object detection, generative models)
- Reinforcement learning (policy optimization, model-based RL, multi-agent)
- Graph learning (GNNs, molecular modeling, knowledge graphs)
- Generative models (diffusion, GANs, VAEs, flow models)
- Causality and interpretability
- AI for science (protein folding, drug discovery, climate modeling)

### Research Setting Adaptations
- Academic research (university, research institution)
- Industry research labs (DeepMind, OpenAI, FAIR, Google Brain)
- Hybrid academic-industry collaborations
- Government research (DARPA, NSF-funded projects)
- Startup R&D (high-risk, high-reward research)
- Open-source AI research (EleutherAI, Hugging Face)

### Publication Venue Focus
- Machine learning (NeurIPS, ICML, ICLR, AAAI)
- Computer vision (CVPR, ICCV, ECCV)
- Natural language processing (ACL, EMNLP, NAACL)
- Artificial intelligence (IJCAI, AAAI)
- Theory (COLT, ALT)
- Interdisciplinary (Nature Machine Intelligence, Science Robotics)

### Career Stage Adaptations
- PhD student (dissertation research, first papers)
- Postdoctoral researcher (establishing research agenda)
- Research scientist (industry labs)
- Assistant professor (tenure-track, lab building)
- Senior researcher (established research program)
- Research director (managing research teams)

## Key Deliverables

1. **Research Publications**
   - Conference papers (NeurIPS, ICML, ICLR, CVPR, ACL)
   - Journal articles (JMLR, PAMI, Nature Machine Intelligence)
   - Workshop papers and extended abstracts
   - ArXiv pre-prints with version tracking
   - Technical reports and whitepapers

2. **Research Artifacts**
   - Open-source code repositories (GitHub)
   - Pre-trained model checkpoints
   - Datasets and benchmarks
   - Reproducibility artifacts (Docker, configs)
   - Interactive demos and visualizations

3. **Theoretical Contributions**
   - Mathematical proofs and derivations
   - Complexity analysis and bounds
   - Convergence guarantees
   - Sample complexity results
   - Generalization theory

4. **Conference Materials**
   - Oral presentation slides
   - Research posters
   - Live demos and visualizations
   - Workshop tutorials
   - Panel participation materials

5. **Community Engagement**
   - Blog posts and explainer articles
   - Video presentations (YouTube)
   - Social media engagement (Twitter threads)
   - Open review participation
   - Mentorship and teaching materials

## Metrics and KPIs

### Publication Impact
- Papers at top-tier venues (target: 2-3 per year for senior researchers)
- Conference acceptance rates (top venues: 20-25%)
- Oral presentation rate (target: 20% of accepted papers)
- Best paper awards and honorable mentions
- Citation counts (target: 50+ citations per paper within 2 years)
- h-index growth (target: +2-3 per year for mid-career)

### Research Contributions
- State-of-the-art (SOTA) results on benchmarks
- Novel theoretical results (proofs, bounds)
- Released open-source implementations (GitHub stars: >500 for impactful work)
- Pre-trained models downloaded (Hugging Face downloads: >10K)
- Benchmark leaderboard positions
- Reproducibility and replication by others

### Research Funding
- Grant success rate (NSF: 20-25%, industry: variable)
- Total funding secured (target: $500K+ per year for PI)
- Number of active grants (target: 2-3 concurrent)
- Collaborative grants and partnerships
- Equipment and compute resource access
- Student and postdoc funding

### Academic Recognition
- Invited talks at universities and conferences (target: 3-5 per year)
- Keynote speeches at major venues
- Best paper awards and nominations
- Outstanding reviewer awards
- Program committee memberships
- Area chair or senior area chair roles
- Editorial board memberships

### Research Impact
- Adoption by industry (products using research)
- Integration into ML frameworks (PyTorch, TensorFlow)
- Media coverage and public engagement
- Policy influence (AI safety, ethics)
- Mentee success (student placements, papers)
- Open-source community engagement
- Research collaborations initiated

### Quality & Rigor
- Reproducibility rate (>90% of results reproducible)
- Code quality and documentation standards
- Experimental rigor (multiple seeds, significance tests)
- Ablation study completeness
- Baseline comparison fairness
- Limitation disclosure transparency
- Peer review scores (target: weak accept or higher)

---

**Note**: This preset provides comprehensive AI research guidance for cutting-edge machine learning and artificial intelligence research. Research directions, methodologies, and publication standards evolve rapidly in AI/ML. Always stay current with latest developments, engage with the research community, consult recent papers from top venues, and adhere to field-specific best practices and ethical guidelines. Specific research directions should be informed by current state-of-the-art, open problems, and available computational resources.
finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the workStars: 323kUpdated Claude CodeCodexVerifiedInstall
systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixesStars: 323kUpdated Claude CodeCodexVerifiedInstall
test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation codeStars: 323kUpdated Claude CodeCodexVerifiedInstall
skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.Stars: 198kUpdated Claude CodeCodexVerifiedInstall